Illustrative · Regulated delivery office
Modeling a sovereign delivery office on Homany
A worked scenario - not a customer result - showing how a 200-person delivery office in a regulated, data-localized market would stand up sovereign work execution with agents, and what the economics look like under usage-based pricing.
This is a modeled scenario, not a customer result. Figures labeled “modeled” are illustrative estimates that depend on your own inputs. We won’t publish a real customer outcome until that customer signs off on the numbers.
The numbers
- Lower collaboration license cost modeled
- Data stays in-region measured
- Modeled agent runs / month modeled
~70%
Modeled: 160 collaborators move from paid seats to free, leaving only 40 editors plus metered AI. Figure depends on your seat price and mix.
100%
Structural, not modeled: the workspace runs in the customer's own cloud account in the required jurisdiction.
~2,000
Assumed volume for status roll-ups, intake triage, and reporting. Cost scales with this number and your token assumptions.
The constraint: AI ambition, sovereign reality
Picture a delivery office coordinating dozens of programs under a data-localization regime. Leadership wants AI assistance on the busywork - status chasing, intake triage, reporting - but the data cannot leave the jurisdiction, and a per-seat tool taxes every reviewer and sponsor who only needs to look.
The usual options force a trade: adopt a SaaS PM tool and fight the residency battle, or stay on spreadsheets and email and forgo agents entirely. This scenario models a third path.
“Residency stops being a negotiation and becomes a deployment setting your security team can see.”
The approach: workspace in-region, collaborators free
Homany deploys into the customer's own cloud account in the required region. The workspace - sheets, views, the execution graph - never leaves that boundary, and the security team can verify where it runs rather than take a slogan on trust.
Editors who build and own work hold the paid roles. The 160 people who view, comment, and request are collaborators, and collaborators are free. AI work is metered in tokens with admin budgets and per-agent caps, so finance can predict and govern spend.
The modeled outcome: predictable cost, agents in-boundary
Under these assumptions, collaboration license cost falls sharply because most participants no longer need a seat, while AI cost becomes a function of run volume the admin team controls.
These are modeled figures, not a measured deployment. The point isn't the exact percentages - it's the structure: residency is a deployment property you can verify, and the economics follow the work the AI actually does, not the size of the audience watching it.